TechSoft
Playbook menu

TechSoft Delivery Playbook

One team. Clear responsibilities. Reliable software.

We combine experienced engineers, AI-powered workflows and rigorous quality controls to help businesses build, modernize and operate software with confidence.

Overview: how we deliver software

The playbook is our written description of how a TechSoft team takes a client's problem to reliable, working software: the stages, the roles, the rules and the weekly rhythm. We publish it so you can judge how we work before you hire us.

The problems it addresses

  • Software that takes too long to ship, or ships with too many surprises.
  • Old systems that are costly and risky to change.
  • AI tools adopted without clear rules for quality, security or ownership.
  • Paying for hours instead of getting a result you can measure.

How AI changes our delivery

AI helps us draft code, tests and documentation, understand old code and investigate defects. That makes writing code cheaper. It does not make it safe or right. What a client should pay for is judgment, discipline and accountability, so that is what we put around the tools.

Why people stay accountable

An AI cannot be held to account. Every change is understood and owned by an engineer, and architecture, security exceptions and releases are approved by named people.

What you can expect

  • The same five stages, with a clear milestone at each.
  • Working software in short, regular increments.
  • One backlog, a weekly status and an honest risk list.
  • Progress measured against a baseline we agree with you.

Built on four values

#Kindness

We treat people the way we want to be treated. We are honest and open, we respect one another, and we support and encourage each other, in the team and with our clients.

#Sustainability

We think long term. We look after our health, keep learning and growing, and share what we know with our team and the wider community.

#One-Team

We share goals and succeed together. Everyone takes ownership of their work, looks for practical ways to move forward, and celebrates results as a team, including our clients.

#Excellence

We care about quality and stand behind our work. We hold ourselves accountable and aim to get a little better every day.

See what each value looks like in daily work →

The delivery lifecycle

Every engagement moves through five stages, scaled to its size. For each one you can see what happens, what you receive, which quality checks apply and what must be approved before we move on.

= milestone agreed with the client

  1. 01

    Discover

    Problem and baseline

  2. 02

    Design

    Architecture and scope

  3. 03

    Build with AI

    Reviewed increments

  4. 04

    Verify

    Gates and acceptance

  5. 05

    Operate & improve

    Monitor and learn

Measure outcomes, learn, and feed the next increment
Each stage ends with a milestone agreed with the client. What we learn in Operate feeds the next round of Discover.
  1. STAGE 01

    Discover

    Understand the business problem, set the baseline and success measures, and surface constraints and risks.

    You receive

    • Problem statement and success measures
    • Current-state map
    • Risk register

    Quality checks

    • Problem statement and success measures read back to the client
    • Baseline numbers recorded, not estimated

    Approved before we move on

    The Client Sponsor agrees the problem statement, success measures and next step.

  2. STAGE 02

    Design

    Define requirements and acceptance criteria, weigh the technical options, and plan the delivery.

    You receive

    • Architecture and ADRs
    • Backlog and roadmap
    • Prototypes and design specs

    Quality checks

    • Architecture reviewed by the Tech Lead; key decisions recorded
    • Every story has testable acceptance criteria

    Approved before we move on

    Architecture, scope for the first release and acceptance criteria approved by client and Tech Lead.

  3. STAGE 03

    Build with AI

    Work in small increments. AI tools and agents help where appropriate; engineers review and own every change.

    You receive

    • Working increments in a shared environment
    • Automated tests
    • Updated documentation

    Quality checks

    • Peer review of every change, including AI-assisted ones
    • Automated tests, static analysis and dependency checks in the pipeline

    Approved before we move on

    Each story meets the definition of done; the increment is demonstrated to the client.

  4. STAGE 04

    Verify

    Run automated and manual tests, check security, reliability and maintainability, and get the approvals needed to release.

    You receive

    • Test report and open-defect list
    • Security findings and decisions
    • Acceptance sign-off

    Quality checks

    • Regression, integration and performance testing
    • Security and dependency checks; findings triaged

    Approved before we move on

    Release gates passed; the Client Sponsor approves the release.

  5. STAGE 05

    Operate and improve

    Monitor production, track incidents, defects and performance, and feed what we learn back into the next cycle.

    You receive

    • Runbooks and dashboards
    • Monthly report
    • Updated roadmap

    Quality checks

    • Monitoring and alerts reviewed
    • Incidents get a post-incident review

    Approved before we move on

    Ongoing: priorities and improvements reviewed monthly with the client.

See who is involved in each stage →

AI engineering, with people in charge

AI supports the whole workflow, from requirements to production incidents. We apply it in three levels, step by step, with defined permissions and human approval.

Level 1 · Where it starts

AI-assisted engineering

Individual engineers use approved AI tools to speed up specific tasks, such as drafting code and tests, explaining unfamiliar code or summarizing requirements. The engineer reviews and owns the result.

Level 2 · Where we are building toward

AI-orchestrated delivery

Controlled AI-agent workflows handle bounded engineering tasks, for example a refactor or a test suite. Each has defined permissions, automated verification and clear rules for when to stop and ask a person.

Level 3 · The longer-term direction

AI-enabled operations

AI helps monitor running systems, diagnose incidents, spot recurring defects and technical debt, and suggest improvements. People decide what changes in production.

These are capabilities and working patterns we apply step by step, agreed with each client. They are not a claim that every capability is already in use on every project.

Who does what

AI

Generates, analyzes, summarizes and recommends

Never approves, accepts or releases anything.

Engineers

Review, validate, understand and own changes

If an engineer cannot explain it, it does not ship.

Automated controls

Test, scan, validate and enforce defined policies

Failing checks block the merge or the release.

Human decision-makers

Approve architecture, security exceptions, releases and consequential production changes

Named people, recorded decisions.

Your code and data

Client code and data

Agreed at kickoff in the engagement's AI-use policy. Only approved tools touch client code or data.

Access control

AI tools and agents get the minimum access they need, on specific repositories and environments, and nothing more.

Secrets and regulated data

Credentials, secrets and regulated data stay out of AI tools unless the policy explicitly allows it.

Auditability

Changes live in version control with the author, the reviewer and the checks that ran. We record when AI assisted a change where the policy requires it.

Human approval

Architecture, security exceptions, releases and consequential production changes are approved by named people.

Tool terms

Each AI tool has its own data-retention and training terms. We check them before approving a tool, and we tell you which tools we use. We do not promise anything beyond what a tool's terms and our controls support.

Read the full AI engineering page →

Quality, security and governance

These controls are what make AI-powered engineering dependable. They reduce risk; they do not promise perfect security or zero defects.

  • Version control and traceable changes

    You can see who changed what, when and why, and go back if something goes wrong.

  • Automated testing

    Mistakes are caught in minutes, not discovered by your users.

  • Code review

    A second pair of human eyes on every change, including AI-assisted ones.

  • Static analysis and dependency scanning

    Common defects and known-vulnerable libraries are flagged before release.

  • Security review and remediation

    Findings are triaged, assigned an owner and tracked until closed or consciously accepted.

  • Acceptance criteria

    Everyone agrees up front what “working” means, so acceptance is not an argument.

  • Release approval

    Nothing reaches production without a named approver.

  • Rollback and recovery planning

    If a release misbehaves, there is a tested way back.

  • Production monitoring

    Problems are seen by us first, with alerts and dashboards.

  • Decision rights and escalation

    Everyone knows who decides what, and how a problem moves up quickly.

See the four quality gates →

Metrics that connect to outcomes

Each measure has a baseline, a measurement period and one agreed definition. These are the measures we propose. Actual results are reported per engagement, and we never use lines of code as a productivity measure.

Delivery

How quickly useful software reaches users.

Lead time for changes
Time from a change being started to it running in production
Time to first usable release
Time from kickoff to the first release real users can use
Deployment frequency
How often we release to production

Quality and reliability

Whether what we ship works and stays working.

Escaped defects
Defects found in production that testing should have caught
Change failure rate
Share of releases that cause a failure or need a fix or rollback
Time to restore service
How long it takes to recover from an incident
Security findings and remediation
Findings raised, and how long they take to close

AI effectiveness

Whether AI is really helping, not just producing more code.

Time saved on selected tasks
Compared with the same kind of task before AI, on agreed examples
AI-assisted changes accepted after review
Share of AI-assisted changes that pass review
Rework from AI-generated changes
Fixes and rewrites needed after AI-assisted work
Verification and review effort
Time people spend checking AI output

Business outcomes

Whether the software moved something the client cares about.

Time to market
Time from idea to clients using it
Feature adoption
How many people actually use what we shipped
Operating cost
Cost to run and support the software
Client-defined return on investment
Measured against the client's own definition of value

Delivery

Faster, smaller releases

Quality and reliability

Fewer failures, faster recovery

AI effectiveness

Time saved, minus the checking it needs

Business outcomes

Time to market, adoption, cost, your own ROI

Engineering measures only matter when they move an outcome you care about. That is why each engagement starts by agreeing the outcome and its baseline.
See how we report →

Three ways to start

You do not need a large commitment to find out whether this works for you.

  1. Path 1

    AI Engineering Assessment

    A focused look at your architecture, development process, AI readiness, technical debt and quality risks.

    Possible deliverables

    • Current-state assessment
    • Prioritized improvement roadmap
    • Recommended pilot scope
    • Baseline metrics
  2. Path 2

    AI Engineering Pilot

    We apply the approach to one bounded feature, workflow or modernization task, so you can judge it on real work.

    Possible deliverables

    • Working software
    • Test and quality evidence
    • Documented workflow improvements
    • Comparison against an agreed baseline
  3. Path 3

    Continuous Product Engineering

    We extend what worked into ongoing engineering, modernization, quality and operations.

    Possible deliverables

    • Prioritized roadmap
    • Regular software releases
    • Quality and operational reporting
    • Continuous improvement recommendations

Scope, timing and commercial terms are agreed with you for each engagement. This page does not set prices or standard durations.

Evidence from real engagements

These engagements are already public on our site. We publish only approved details and results.

How we document our case studies →

Detailed working standards

The reference material our teams work from. Open a group to see its pages. Every page names an owner and the date it was last reviewed, and is reviewed again at least every six months.

Client

  • Client Sponsor
  • Product owner and technical contacts
  • Users and stakeholders

TechSoft delivery team

Leadership

  • Delivery Manager
  • CTO (security, compliance, AI and data)

Design

  • UX/UI Designer
  • Requirements work is shared across the whole team

Engineering

  • Solution Architect / Tech Lead
  • Software Engineers

Quality & operations

  • QA / Test Automation
  • DevOps / Platform
  • Support & Operations

One backlog · one definition of done · one status report

All roles work from the same plan, visible to the client

The client and the team work together every week.

Small engagements combine several roles in one person, but every responsibility is still covered.
Start here 2 pages
Roles 4 pages
Lifecycle 6 pages
Working agreements 3 pages
Quality & AI 4 pages
Craft 3 pages
Services 1 pages
Governance 2 pages
People and careers 5 pages